Eyewitness identification accuracy from point-light lineups

Abstract

By using the point-light technique to isolate kinematic information, Cutting and Kozlowski (1977) found that people can recognise familiar others from their gait. This technique involves attaching lights to individuals' joints, filming them engaged in activity, and altering the contrast on the monitor to give a display of moving lights. This thesis describes a series of experiments that used a modification of Cutting and Kozlowski's method to investigate whether people can identify targets seen only once, from their kinematic information. In Experiment 1 five subjects, all familiar with each other, could identify each other at a rate significantly higher than chance. Moreover, there was a positive relationship between identification accuracy and confidence. In Experiment 2, 48 subjects viewed a 5 minute video of a moving target person, and subsequently tried to identify the target from either a point-light or control lineup. Control identification was significantly higher than chance, but point-light identification was no different from chance. Control subjects used body shape information in addition to movement cues to aid their identification, a behaviour that motivated a further experiment. Experiment 3 tested whether subjects would be more reliant on movement information if they viewed the target under reduced lighting conditions that hid other information such as body shape. Overall, point-light subjects' identification accuracy was not significantly higher than chance; however, it was significantly higher than chance for identifications of female targets. Results are discussed with respect to Runeson's (1985) KSD-Principle, interpreted from an ecological viewpoint, and related to implications for eyewitness identification procedures.